JP2017169022A - Moving image encoding device, method, program, and moving image encoding system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a moving image encoding device that does not require a cache mechanism to store a moving image and does not cause delay.SOLUTION: A moving image encoding device 100 comprises: a reference image reading unit that reads image data of a predetermined area of a reference image corresponding to an encoding object image from an external storage device, and stores it in an internal buffer; and a vector search unit that generates a plurality of motion vectors by performing motion search on a plurality of search ranges having different ranges using the image data stored in the internal buffer, and outputs the best motion vector as an evaluation motion vector using an evaluation function. The reference image reading unit reads from the external storage device the image data using an area that has at least two times of longitudinal and lateral length of the search area as the predetermined area compared to the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector where the vector search unit performs the motion search, and stores it in the internal buffer.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a moving image encoding apparatus, method, program, and moving image encoding system.

  As a method of compressing a moving image, a method of encoding a moving image using a similarity between a plurality of frames constituting a moving image (each still image forming a moving image) is known. In this method, for example, a difference between a reference still image (hereinafter referred to as a reference image) before and after (past or future) before and after a reference still image (hereinafter referred to as an encoding target image) is calculated. Encode a video. In general, when encoding a moving image, each still image is divided into a plurality of regions in units of blocks, and pixel data for each block is compared for similarities between the encoding target image and the reference image. To generate a predicted image.

  A method for encoding a moving image will be described with reference to FIG. FIG. 10 is a conceptual diagram for explaining a method of encoding a moving image. FIG. 10A shows a method of encoding a moving image without using a reference image, and FIG. 10B shows a reference image. 3 shows a method of encoding a moving image using.

  In FIG. 10A, consider a case where a block 11 (hereinafter referred to as an encoding target block) surrounded by a square in the center of the encoding target image 10 is encoded. If the encoding target block 11 is encoded as it is, the amount of code increases.

  On the other hand, as shown in FIG. 10B, a method of encoding the encoding target block 11 using the reference image 12 is considered. In this case, first, a frame before and after the encoding target image 10 is defined as a reference image 12, and a block of pixel data similar to the encoding target block 11 (hereinafter referred to as a prediction target block) is searched from the reference image (hereinafter referred to as a prediction target block 11). , Called motion search). And the prediction image 13 containing the prediction object block 14 is produced | generated based on the search result of a motion search.

  Next, the encoding target image is encoded based on the difference value between the prediction target block 14 and the encoding target block 11. At this time, the final code amount is a value obtained by adding the difference value between the predicted image and the reference image to the motion vector indicating the position of the reference image.

  Since the difference between the encoding target image and the prediction image similar to the encoding target image is small, the encoding method using the reference image has a code amount as compared with the case of encoding without using the reference image. Is small.

  However, the method of encoding a moving image using a reference image needs to perform a motion search for the entire range of the reference image. Since it takes time to perform motion search for the entire range of the reference image, it is difficult to encode the moving image using the reference image in an encoder, hardware encoder, or the like that needs to be encoded at high speed. It is. For this reason, for motion search, a method of limiting the range that a motion vector can take has been studied. Hereinafter, the range that the motion vector can take is referred to as a search permission range. Also, instead of searching all within the search-permitted range, the search is limited to a certain range from the center of the search-permitted range in order to search only a range where there is a high possibility that similarities with the encoding target block exist. The technique to do is also examined. Hereinafter, a range in which motion search is actually executed is referred to as a search range.

  The moving image encoding device described in Patent Document 1 determines the center of motion search by determining the similarity of surrounding images in the same frame as the encoded image, and optimizes the search range of the motion vector. .

  The motion search described in Non-Patent Document 1 limits the search range of motion vectors by selecting and deriving the center of motion search from a plurality of candidates.

JP 2006-166247 A

L.-F. Ding, W.-Y. Chen, P.-K. Tsung, T.-D. Chuang, P.-H. Hsiao, Y.-H. Chen, H.-K. Chiu, S .-Y. Chien, and L.-G. Chen, “A 212 Mpixels / s 4096x2160p multiview video encoder chip for 3D / quad full HDTV applications,” IEEE Journal of Solid-State Circuits, vol. 45, no. 1, pp. 46-58, Jan. 2010.

  When encoding a moving image having a large screen size such as Full HD (High Definition Video), 4K, or 8K, even if a technique for limiting the search range of motion search to a certain range from the search center is used, The reference location varies depending on the block. Therefore, in order to encode a moving image at high speed, it is necessary to store image data in the search range in the apparatus. However, when the search permission range is large, the amount of information becomes enormous, so it is difficult to install a storage device that can store all the image data in the search permission range. Therefore, when the image data is large, the image data is usually stored in an external storage device. Here, in the case of a hardware encoder, for example, the external storage device is an SDRAM (Synchronous Dynamic Random Access Memory) installed outside the encoder core. However, when image data is stored in the external storage device, the external storage device has a large access delay, and it takes a long time to read the image data.

  Patent Document 1 includes an encoded image memory that outputs an encoded image and a reference image memory that outputs a reference image. Therefore, in Patent Document 1, when the screen size becomes large, it is necessary to store image data necessary for encoding in an external storage device, and therefore there is a possibility that a delay occurs in encoding.

  Non-Patent Document 1 employs a method in which all data is not retained by a cache mechanism, but the cache mechanism becomes complicated, and the hardware encoder increases the circuit scale or causes a delay when there is no cache hit. there is a possibility.

  An object of the present invention is to provide a moving image encoding apparatus, method, program, and moving image encoding system that do not require a cache mechanism for holding a moving image and that do not cause a delay when encoding a moving image. There is to do.

  The moving image encoding device according to one aspect of the present invention reads image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting the moving image, Using the reference image reading unit stored in the internal buffer and the image data of the predetermined region of the reference image stored in the internal buffer, the ranges differ from the encoding target region of the encoding target image, respectively. A vector search unit that performs a motion search for first to Nth (N is an integer of 2 or more) search ranges, and generates search results as first to Nth motion vectors, respectively. Compares the evaluation function values of the first to Nth motion vectors, selects the motion vector having the best evaluation function value, outputs it as an evaluation motion vector, and reads the reference image And a search unit for the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector for which the vector search unit performs a motion search from the external storage device, The image data is read out with the area obtained by multiplying at least twice as the predetermined area and stored in the internal buffer.

  The moving image encoding method according to another aspect of the present invention reads image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting the moving image. , Stored in the internal buffer, and using the image data of the predetermined area of the reference image stored in the internal buffer, the first to the first different from the encoding target area of the encoding target image, respectively N (N is an integer greater than or equal to 2) search range, search results are generated as first to Nth motion vectors, and evaluation function values of the first to Nth motion vectors are obtained. The motion vector having the best evaluation function value is selected and output as an evaluation motion vector, and the search range corresponding to the evaluation motion vector and the vector search unit move from the external storage device. The image data is read out with the search area corresponding to the motion vector on which the search is performed at least twice as long as the vertical and horizontal of the search range as the predetermined area, and stored in the internal buffer.

  According to another aspect of the present invention, there is provided a moving image encoding program that stores an image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting the moving image. Using the reference image reading means for storing the image in the internal buffer and the image data of the predetermined area of the reference image stored in the internal buffer, with respect to the encoding target area of the encoding target image, A vector search means for executing a motion search for first to Nth (N is an integer of 2 or more) search ranges having different ranges, and generating search results as first to Nth motion vectors, respectively, Means for comparing evaluation function values of the Nth motion vector, selecting a motion vector having the best evaluation function value, and outputting it as an evaluation motion vector; and the external storage From the device, a search range corresponding to the evaluation motion vector and a search range corresponding to a motion vector for which the vector search unit performs a motion search are obtained by multiplying the vertical and horizontal directions of the search range by at least twice. The image data is read out as the predetermined area and functions as means for storing in the internal buffer.

  A moving image encoding system according to another aspect of the present invention includes an external storage device that holds a plurality of still images constituting a moving image, and a predetermined region of a reference image corresponding to the encoding target image from the external storage device. A reference image reading unit that reads image data and stores the image data in an internal buffer, and the image data of a predetermined region of the reference image stored in the internal buffer is used to encode the encoding target region of the encoding target image. A vector search unit that performs a motion search for first to Nth (N is an integer of 2 or more) search ranges having different ranges, and generates search results as first to Nth motion vectors, respectively. The vector search unit compares the evaluation function values of the first to Nth motion vectors, selects a motion vector having the best evaluation function value, and outputs it as an evaluation motion vector. The reference image reading unit searches the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector on which the vector search unit performs a motion search from the external storage device. The image data is read out with the area obtained by at least doubling the length and width of the range as the predetermined area, and stored in the internal buffer.

  According to the present invention, there is provided a moving image encoding apparatus, method, program, and moving image encoding system that do not require a cache mechanism for holding a moving image and that do not cause a delay when encoding a moving image. can do.

It is a block diagram which shows the structure of the moving image encoder which concerns on the 1st Embodiment of this invention. It is a conceptual diagram for demonstrating the concept of the motion vector which concerns on this invention. It is a schematic diagram for demonstrating operation | movement of the moving image encoder which concerns on the 1st Embodiment of this invention. It is a schematic diagram for demonstrating the search range of the motion search which concerns on this invention. It is a flowchart which shows the flow of operation | movement of the moving image encoder which concerns on the 1st Embodiment of this invention. It is a block diagram which shows the structure of the moving image encoder which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the flow of operation | movement of the moving image encoder which concerns on the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the moving image encoding system which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structure of the information processing apparatus which comprises the moving image encoder which concerns on this invention. It is a conceptual diagram for describing a method of encoding a moving image, (a) shows a method of encoding a moving image without using a reference image, and (b) encodes a moving image using a reference image. Shows how to

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In addition, in order to avoid the complexity by having demonstrated repeatedly, the same code | symbol is attached | subjected to the part which is the same or it corresponds in each figure, and description is abbreviate | omitted suitably.

[First Embodiment]
[Moving picture encoding device]
With reference to FIG. 1, a video encoding apparatus 100 according to the first embodiment of the present invention will be described.

  The moving image encoding apparatus 100 includes a reference image reading unit 110 and a vector search unit 120.

  The reference image reading unit 110 reads the reference image of the encoding target image from an external storage device (not shown) that holds a moving image composed of a plurality of still images (frames). The reference image means a past or future still image with respect to the encoding target image. The reference image reading unit 110 stores the reference image read from the external storage device in the internal buffer. Here, the internal buffer may be included in the reference image reading unit 110, or may be included in the video encoding device 100 independently of the reference image reading unit 110.

  Based on the encoding target image and the reference image stored in the internal buffer, the vector search unit 120 performs a motion search in an arbitrary search range on the encoding target region of the encoding target image, and moves the search result to the motion Generate as a vector. Here, the vector search unit 120 may specify the center of the search range by position coordinates, or may specify the vector from the encoding target region. Further, the vector search unit 120 executes a motion search for each search range when any N (N is an integer of 2 or more) search ranges having different ranges are set as the first to Nth search ranges. Then, the search result is generated as the first to Nth motion vectors. Further, the vector search unit 120 compares the evaluation function values of the first to Nth motion vectors, selects one motion vector having the best evaluation function value, and outputs it as an evaluation motion vector.

  Here, as the evaluation function value, for example, SAD (Sum of Absolute Difference) for searching for a point having a high similarity between the encoding target image and the reference image can be used. In this case, the best motion vector means a motion vector having the smallest SAD value. Note that the method of comparing the evaluation function values is not limited to the above, and any evaluation method that can compare the encoding target image 20 and the reference image 30 may be used.

[Concept of motion vector]
With reference to FIG. 2, the concept of motion vectors generated by the video encoding device 100 will be described. FIG. 2 shows an encoding target image 20 and a reference image 30.

  The encoding target image 20 is an arbitrary still image serving as a reference when encoding a moving image among a plurality of still images constituting the moving image. In FIG. 2, the encoding target image 20 includes an encoding target block 21. The encoding target block 21 is an area to be encoded in the encoding target image 20.

  The reference image 30 is, for example, one frame of past (or future) still images with respect to the encoding target image 20 among the plurality of still images constituting the moving image. The search range 31 is a range in which the moving image coding apparatus 100 performs a motion search in the reference image 30. The search range 31 is not particularly limited, but is preferably in the vicinity of the encoding target block 21. This is because when the reference image 30 is a still image of one frame past the encoding target image 20, the position of the encoding target block 21 is not likely to change significantly. Further, the motion search may be executed at the same position as the encoding target image block 21 in the reference image 30 by setting the range in which the motion search is executed as the zero vector of the encoding target block 21. The prediction target block 32 is an area most similar to the encoding target block 21 in the search range 31. That is, the video encoding device 100 searches the position of the prediction target block 32 by searching the search range 31.

  The motion vector 40 indicates the motion of the encoding target block 21 with respect to the prediction target block 32. In FIG. 2, only the motion vector 40 is shown as the motion vector. However, the moving image encoding apparatus 100 can generate a plurality of motion vectors by setting a plurality of ranges in which motion search is executed.

[Operation of Video Encoding Device]
A specific example of the operation of the moving image encoding apparatus 100 will be described with reference to FIG. FIG. 3 shows an encoding target image 20 ′ having a plurality of areas and a reference image 30 ′.

  As shown in FIG. 3, the encoding target image 20 ′ includes a total of 12 areas of 4 blocks horizontally × 3 blocks vertically 1 to 12. The reference image 30 ′ includes a total of 12 areas of blocks 1 ′ to 12 ′. Here, the positions of the blocks 1 'to 12' correspond to the positions of the blocks 1 to 12, respectively.

  In order to generate a motion vector of an encoding target block among the blocks 1 to 12, it is effective to execute a motion search using a block near the encoding target block as a search range. For example, when the block to be encoded is block 2, when generating the motion vector of block 2, a motion search may be executed using the vector (position) of block 1 ′ adjacent to the left as a search range. It is effective. In this case, the moving image encoding apparatus 100 reads out the block 1 'as the search range from the reference image 30' from the external storage device, and executes a motion search. If a motion vector cannot be generated by the motion search of block 1 ′, the moving image coding apparatus 100 performs horizontal operation in the order of block 2 ′, block 3 ′, and block 4 ′ after the search of block 1 ′. Assume that motion search is performed on a row of blocks.

  Here, when the motion search is performed with respect to the vector of block 1 ′ and the motion search is performed for the next block 2, the moving picture decoding apparatus 100 uses the external storage device after the motion search of block 1 ′ is completed. As a range of motion search, block 2 ′ is read out as a reference image. For this reason, since the moving picture coding apparatus 100 performs the motion search of the block 2 after reading the reference image from the external storage device, the time until the reference image is read from the external storage device is a delay time, and the coding efficiency is Getting worse. For example, when the block to be encoded is block 6 and the motion search is performed on the upper adjacent block 2 ′, the same as when the motion search is performed on the horizontal row of blocks. If there is a delay, a delay occurs until the motion search is executed.

  For this reason, the moving picture encoding apparatus 100 according to the first embodiment eliminates the delay time by prefetching information on motion vectors of adjacent blocks two blocks to the left of the encoding target block, for example. Specifically, in the first embodiment, the reference image reading unit 110 can use the motion search result of the vector search unit 120 of the video encoding device 100 to eliminate the delay time.

  Below, the moving image encoding apparatus 100 demonstrates the method of eliminating delay time by performing a motion search about two search ranges with respect to an encoding object block.

  The two search ranges are not particularly limited. For example, a search using a vector (position) of the block adjacent to the left of the encoding target block as the center of the motion search and a vector different from the vector of the block adjacent to the left ( Motion search centered on (position). Here, the vector different from the vector of the block adjacent to the left of the encoding target block is not particularly limited, but may be a zero vector of the encoding target vector, for example. Further, as a vector different from the motion vector of the block adjacent to the left, for example, a motion vector obtained by searching a wide range in advance using a reduced image or an image obtained by thinning out pixels may be used.

  In addition, although the case where a moving image is encoded based on two motion vectors for simplicity will be described, this is an exemplification and does not limit the present invention. The present invention can encode an encoding target image based on N motion vectors.

  First, the reference image reading unit 110 reads the reference image corresponding to the encoding target image from the external storage device and stores it in the internal buffer.

  Based on the encoding target image and the reference image stored in the internal buffer, the vector search unit 120 performs a motion search centering on a vector of a block adjacent to the left of the encoding target block, and obtains the search result as a first result. Generate as a motion vector. Here, the vector search unit 120 performs two types of motion searches when generating the first motion vector. Specifically, the motion search is centered on the zero vector of a block adjacent to the left of the encoding target block, and the motion search is centered on the vector of the adjacent block two blocks to the left of the encoding target block.

  Moreover, the vector search part 120 performs a motion search centering on the zero vector of an encoding object block, and outputs a search result as a 2nd motion vector. Then, the vector search unit 120 compares the evaluation function value of the first motion vector with the evaluation function value of the second motion vector, and outputs the better evaluation function value as an evaluation motion vector. Here, the evaluation motion vector is, for example, the motion vector of the adjacent block two blocks to the left of the encoding target block. Also, a block in the reference image specified by the evaluation motion vector is set as a prediction target block. Further, the vector search unit 120 searches for a motion in a search range different from the search range corresponding to the first motion vector and the second motion vector, and generates a third motion vector.

  The reference image reading unit 110 reads from the external storage device, for each of the evaluation motion vector and the third motion vector, a reference image in which an area at least twice as long and horizontal as the search range in the vector search unit 120 is added. Then, the reference image reading unit 120 stores the read reference image in the internal buffer.

  Next, the vector search unit 120 performs a motion search for the prediction target block. Here, the vector search unit 120 performs a motion search for two search ranges when generating a motion vector of a block adjacent to the left of the encoding target block. Therefore, the vector search unit 120 performs a motion search for the three search ranges for the prediction target block.

  The first is a motion search (hereinafter referred to as search A) using the prediction target block as a zero vector. The second is a motion search (hereinafter referred to as search B) centered on a motion vector generated by a motion search centered on a zero vector of a block adjacent to the left of the encoding target block. The third one is a motion search (hereinafter referred to as search C) centered on a motion vector generated by a motion search centered on a vector of a neighboring block two left to the encoding target block.

  The reference image reading unit 110 reads a reference image in which an area obtained by adding at least twice the length and width of the search range to the search range centered on the zero vector of the block adjacent to the left of the encoding target block is read. (Hereinafter referred to as read A). In addition, the reference image reading unit 120 reads a reference image to which a search range is added with respect to a search result (prediction target block) centered on the motion vector of the adjacent block two blocks to the left of the encoding target block ( Hereinafter, this is referred to as read B).

  As described above, the vector search unit 120 executes the search C before the reading B. That is, the range of the reference image read by the read B is included in the range of the reference image read by the read A. That is, the reference image reading unit 120 can omit the process of reading B by adding an additional region to the motion vector search result of the adjacent block two blocks to the left of the encoding target block.

  Therefore, the moving image coding apparatus 100 does not need to read the reference image for the search range of the prediction target block from the external storage device when performing the motion search of the prediction target block. Can be resolved.

  FIG. 4 is a schematic diagram showing the relationship between the search range in the vector search unit 110 and the area read out in the reference image reading unit 120. Hereinafter, a method in which the reference image reading unit 120 pre-reads the reference image will be specifically described with reference to FIG.

  The reference image 50 includes an adjacent block 51 of the encoding target block. The adjacent block 51 can be designated by a motion search centered on the zero vector of the adjacent block 51 or the vector of the adjacent block two blocks to the left of the encoding target block. A search range 52 indicates a search range of the adjacent block 51.

  The reference image 60 shows the result of performing a motion search on the search range 52 in the reference image 50, and includes a prediction target block 53. It shows that the prediction target block 53 is located at the lower right of the search range 52.

  The reference image 70 includes an area obtained by doubling at least the vertical and horizontal search ranges with respect to the search range centered on the zero vector of the adjacent block 51 with respect to the reference image 60, and the search range of the prediction target block 53 The result of adding is shown. Here, the search range 54 is a search range for motion search of the prediction target block 53, and the additional region 55 is a region in which the vertical and horizontal are doubled with respect to the search range 52 of the adjacent block 51.

  The search range of the search A described above corresponds to the zero vector of the adjacent block 51 in FIG. That is, the search range of the search A is included in the search range 52 and thus is within the range of the read A. Further, the search range of the search B described above corresponds to the search range 52 in FIG. That is, the search range of search B is within the range of read A as with search A. Further, the search C described above corresponds to the prediction target block 53 in FIG. 4 because it is a motion search centered on a motion vector generated by a motion search centered on the vector of the adjacent block two blocks to the left of the encoding target block. . Therefore, the search range of search C corresponds to search range 54. Here, the search range 54 is within the range of the additional region 55. That is, the search range of the search C is included in the range of the additional region 55 and thus is the range of the read B.

  Therefore, the moving image encoding apparatus 100 reads a reference image including the search range of the prediction target block 53 in advance before reading the reference image of the search range of the prediction target block 53. For this reason, the moving image encoding apparatus 100 can eliminate the time for reading the reference image of the search range of the prediction target block 53, and thus can eliminate the delay in encoding the encoding target image.

  FIG. 5 is a flowchart showing the flow of the operation of the moving picture coding apparatus 10 according to the first embodiment of the present invention. Hereinafter, with reference to FIG. 5, an operation flow of the moving picture encoding apparatus 100 according to the first embodiment of the present invention will be described.

  First, the reference image reading unit 110 reads image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image (step S101). Here, the reference image search range may be specified by the reference image reading unit 110, or may be freely specified by the user of the moving picture coding apparatus 100 or the like.

  Next, the reference image reading unit 110 stores the reference image read from the external storage device in the internal buffer (step S102). Here, in step S101 and step S102, by storing the image data of a predetermined area of the reference image corresponding to the encoding target image in the internal buffer, it is not necessary to refer to the external storage device at the time of encoding. So the delay can be eliminated.

  Next, the vector search unit 120 searches for a search range of a predetermined area of the reference image read by the reference image reading unit 110 in step S102, and calculates a motion vector for each search range (step S103).

  Next, the vector search unit 120 calculates an evaluation function for each of the motion vectors calculated for each search range, and compares the evaluation functions (step S104).

  Next, the vector search unit 120 selects a motion vector of the best evaluation function as a result of the comparison, and outputs the selected motion vector as an evaluation motion vector (step S105).

  Then, the reference image reading unit 110 reads from the external storage device a reference image in which an area at least twice as long and wide as the search range is added to the search range corresponding to the evaluation motion vector (step S106).

  Finally, the reference image reading unit 110 stores the reference image read in step S106 in the internal buffer (step S107).

[Second Embodiment]
FIG. 6 is a block diagram showing a moving picture coding apparatus according to the second embodiment of the present invention.

  The moving image encoding apparatus 100A according to the second embodiment includes a reference image reading unit 110, a vector search unit 120, and a vector setting unit 130. That is, the moving image encoding apparatus 100A is different from the moving image encoding apparatus 100 in that it includes a vector setting unit 130.

  The vector setting unit 130 generates a motion vector different from the first to Nth motion vectors generated by the vector search unit 120 as the set motion vector. Here, the method of generating the set motion vector of the vector setting unit 130 may be a simpler method than the method of generating the first to Nth motion vectors by the vector search unit 120. The vector setting unit 130 generates a set motion vector by searching a wide range of the screen more roughly than the vector search unit 120 by a method such as thinning out a reduced image or pixels of the image, for example. Further, the vector setting unit 130 may generate a specific motion vector corresponding to the position of the screen as a set motion vector based on a feature amount that differs for each coordinate of the still image obtained by image analysis. In this case, the vector search unit 120 compares the evaluation function value of the first to Nth motion vectors with the evaluation function value of the set motion vector, and selects one motion vector having the best evaluation function value. Output as an evaluation motion vector.

  FIG. 7 is a flowchart showing an operation flow of the moving image encoding apparatus 100A according to the second embodiment of the present invention. Hereinafter, with reference to FIG. 7, an operation flow of the moving picture coding apparatus 100A according to the second embodiment of the present invention will be described.

  First, the reference image reading unit 110 reads image data of a predetermined area of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image (step S201).

  Next, the reference image reading unit 110 stores the reference image read from the external storage device in the internal buffer (step S202).

  Next, the vector search unit 120 searches for a search range of a predetermined area of the reference image read by the reference image reading unit 110 in step S202, and calculates a motion vector for each search range (step S203).

  Next, the vector setting unit 130 generates a motion vector different from the motion vector generated by the vector search unit 120 in step S203 as a set motion vector (step S204).

  Next, the vector search unit 120 calculates an evaluation function for each of the motion vectors calculated for each search range in step S203 and the set motion vector, and compares the calculated evaluation function (step S205).

  Next, the vector search unit 120 selects a motion vector of the best evaluation function as a result of comparison, and outputs the selected motion vector as an evaluation motion vector (step S206).

  Then, the reference image reading unit 110 reads, as a reference image, from the external storage device, an area that is at least twice the length and width of the search range in the vector search unit 110 as the search range corresponding to the evaluation motion vector (step S207).

  Finally, the reference image reading unit 110 stores the reference image read in step S207 in the internal buffer (step S208).

[Video coding system]
FIG. 8 is a schematic diagram showing a configuration of a moving picture coding system according to the third embodiment of the present invention.

  As illustrated in FIG. 8, the moving image encoding system 300 includes a moving image encoding device 100 and an external storage device 200.

  The external storage device 200 is not particularly limited, but can be configured by, for example, an HDD (Hard Disk Drive) or the like that is provided inside or outside the moving image encoding device 100. Further, when the moving image encoding apparatus 100 is connected to a communication network such as the Internet, the external storage device 200 may be a cloud storage that exists on the communication network, for example.

  Note that the operation of the moving picture coding system 300 is the same as that of the moving picture coding apparatus 100, and thus the description thereof is omitted. The moving image encoding system 300 may be configured by combining the moving image encoding device 100A and the external storage device 200.

[Other Embodiments]
The moving picture decoding apparatus 100 (FIG. 1) and the moving picture decoding apparatus 100A (FIG. 6) may be realized by hardware or software. Further, the moving picture coding apparatus 100 and the moving picture coding apparatus 100A may be realized by a combination of hardware and software.

  FIG. 9 is an example of an information processing apparatus (computer) constituting the moving picture coding apparatus 100 (FIG. 1) and the moving picture coding apparatus 100A (FIG. 6).

  As illustrated in FIG. 9, the information processing apparatus 400 includes a control unit 401, a storage device 402, a ROM (Read Only Memory) 403, a RAM (Random Access Memory) 404, and a communication interface 405.

  The control unit 401 can be configured by an arithmetic processing device such as a CPU (Central Processing Unit). The control unit 401 configures the moving image encoding device 100 and the moving image encoding device 100A by developing a program readable by the control unit 401 stored in the storage device 402 or the ROM 403 in the RAM 404 and executing the program. Each part can be realized. The control unit 401 may include an internal buffer that can temporarily store data and the like.

  The storage device 402 is a large-capacity storage medium that can hold various data, and can be realized by a storage medium such as a magneto-optical disk, an HDD (Hard Disc Drive), and an SSD (Solid State Drive). The storage device 402 may be a cloud storage that exists on the communication network when the information processing device 400 is connected to the communication network via the communication interface 405. Further, the storage device 401 may hold a program that can be read by the control unit 401.

  The ROM 403 is a non-volatile storage device that can be configured with a flash memory or the like that has a smaller capacity than the storage device 401. The ROM 401 may hold a program that can be read by the control unit 401. Note that the program that can be read by the control unit 401 may be held by at least one of the storage device 402 and the ROM 403.

  The RAM 404 is a semiconductor memory such as a DRAM (Dynamic Random Access Memory) and an SRAM (Static Random Access Memory), and can be used as an internal buffer for temporarily storing data and the like.

  The communication interface 405 is an interface that connects the information processing apparatus 400 and a communication network via a wired or wireless connection.

  Part or all of the above embodiments can be described as in the following supplementary notes. Note that the following supplementary notes do not limit the present invention.

[Appendix 1]
A reference image reading unit that reads out image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image, and stores the image data in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. An integer) search range, and a vector search unit that generates search results as first to Nth motion vectors, respectively,
The vector search unit compares the evaluation function values of the first to Nth motion vectors, selects a motion vector having the best evaluation function value, and outputs it as an evaluation motion vector,
The reference image reading unit performs a vertical search of a search range corresponding to the evaluation motion vector and a search range corresponding to a motion vector on which the vector search unit performs a motion search from the external storage device. A video encoding device that reads out the image data using an area obtained by multiplying at least twice the horizontal and horizontal as the predetermined area and stores the image data in the internal buffer.

[Appendix 2]
The moving picture encoding apparatus according to attachment 1, further comprising a vector setting unit that generates a set motion vector in advance as one of the first to Nth motion vectors.

[Appendix 3]
The vector setting unit according to claim 2, wherein the vector setting unit extracts a feature amount for each coordinate of the reference image based on the image data of the predetermined region, and sets the set motion vector according to the feature amount. Video encoding device.

[Appendix 4]
Image data of a predetermined area of the reference image corresponding to the encoding target image is read from an external storage device that holds a plurality of still images constituting a moving image, and is stored in an internal buffer.
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. (Integer) search range is performed, and search results are generated as first to Nth motion vectors,
Comparing the evaluation function values of the first to Nth motion vectors, selecting the motion vector having the best evaluation function value, and outputting it as an evaluation motion vector;
From the external storage device, the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector for which the vector search unit performs motion search are at least twice the vertical and horizontal directions of the search range. A moving image encoding method, wherein the image data is read out using the determined area as the predetermined area and stored in the internal buffer.

[Appendix 5]
The moving image encoding method according to appendix 4, wherein a set motion vector is generated in advance as one of the first to Nth motion vectors.

[Appendix 6]
6. The moving image encoding method according to appendix 5, wherein a feature amount for each coordinate of the reference image is extracted based on the image data of the predetermined region, and the set motion vector is set according to the feature amount.

[Appendix 7]
Computer
A reference image reading unit that reads out image data of a predetermined area of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image, and stores the reference image in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. A vector search means for executing a motion search for a search range of (integer) and generating search results as first to Nth motion vectors,
Means for comparing evaluation function values of the first to Nth motion vectors, selecting a motion vector having the best evaluation function value, and outputting as an evaluation motion vector;
From the external storage device, the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector for which the vector search unit performs motion search are at least twice the vertical and horizontal directions of the search range. A moving picture encoding program that causes the image data to be read as a predetermined area and to be stored in the internal buffer.

[Appendix 8]
The moving image encoding program according to appendix 7, further causing the computer to function as vector setting means for generating a set motion vector in advance as one of the first to Nth motion vectors.

[Appendix 9]
The computer further extracts a feature amount for each coordinate of the reference image based on the image data of the predetermined area, and further functions as means for setting the set motion vector according to the feature amount. The moving image encoding program described in 1.

[Appendix 10]
An external storage device for holding a plurality of still images constituting a moving image;
A reference image reading unit that reads image data of a predetermined area of a reference image corresponding to an encoding target image from the external storage device, and stores the image data in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. An integer) search range, and a vector search unit that generates search results as first to Nth motion vectors, respectively,
The vector search unit compares the evaluation function values of the first to Nth motion vectors, selects a motion vector having the best evaluation function value, and outputs it as an evaluation motion vector,
The reference image reading unit performs a vertical search of a search range corresponding to the evaluation motion vector and a search range corresponding to a motion vector on which the vector search unit performs a motion search from the external storage device. A moving picture coding system that reads out the image data by using an area obtained by multiplying the horizontal and horizontal at least twice as the predetermined area and stores the read image data in the internal buffer.

[Appendix 11]
The moving image coding system according to appendix 10, further comprising a vector setting unit that generates a set motion vector in advance as one of the first to Nth motion vectors.

[Appendix 12]
The vector setting unit according to claim 11, wherein the vector setting unit extracts a feature amount for each coordinate of the reference image based on the image data of the predetermined region, and sets the set motion vector according to the feature amount. Video coding system.

DESCRIPTION OF SYMBOLS 10 ... Encoding target image 11 ... Encoding target block 12 ... Reference image 13 ... Predicted image 14 ... Prediction target block 20, 20 '... Encoding target image 21 ... Encoding target block 30, 30 '... Reference image 31 ... Search range 32 ... Prediction target block 40 ... Motion vector 50, 60, 70 ... Reference image 51 ... Neighboring block 52 ··· Search range 53 ··· Prediction target block 54 ··· Search range 55 ··· Additional region 100, 100A ··· moving picture encoding device 110 ··· reference image reading unit 120 ··· vector search unit 130 ... Vector setting unit 200 ... External storage device 300 ... Moving picture encoding system 400 ... Information processing device 401 ... Control unit 402 ... Storage device 403 ... ROM
404 ... RAM
405: Communication interface

Claims (10)

A reference image reading unit that reads out image data of a predetermined region of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image, and stores the image data in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. An integer) search range, and a vector search unit that generates search results as first to Nth motion vectors, respectively,
The vector search unit compares the evaluation function values of the first to Nth motion vectors, selects a motion vector having the best evaluation function value, and outputs it as an evaluation motion vector,
The reference image reading unit performs a vertical search of a search range corresponding to the evaluation motion vector and a search range corresponding to a motion vector on which the vector search unit performs a motion search from the external storage device. A video encoding device that reads out the image data using an area obtained by multiplying at least twice the horizontal and horizontal as the predetermined area and stores the image data in the internal buffer.   The moving image encoding apparatus according to claim 1, further comprising a vector setting unit that generates a set motion vector in advance as one of the first to Nth motion vectors.   The vector setting unit extracts a feature amount for each coordinate of the reference image based on the image data of the predetermined area, and sets the set motion vector according to the feature amount. Video encoding device. Image data of a predetermined area of the reference image corresponding to the encoding target image is read from an external storage device that holds a plurality of still images constituting a moving image, and is stored in an internal buffer.
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. (Integer) search range is performed, and search results are generated as first to Nth motion vectors,
Comparing the evaluation function values of the first to Nth motion vectors, selecting the motion vector having the best evaluation function value, and outputting it as an evaluation motion vector;
From the external storage device, the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector for which the vector search unit performs motion search are at least twice the vertical and horizontal directions of the search range. A moving image encoding method, wherein the image data is read out using the determined area as the predetermined area and stored in the internal buffer.   The moving image encoding method according to claim 4, wherein a set motion vector is generated in advance as one of the first to Nth motion vectors.   6. The moving picture encoding method according to claim 5, wherein a feature quantity for each coordinate of the reference image is extracted based on the image data of the predetermined area, and the set motion vector is set according to the feature quantity. . Computer
A reference image reading unit that reads out image data of a predetermined area of a reference image corresponding to an encoding target image from an external storage device that holds a plurality of still images constituting a moving image, and stores the reference image in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. A vector search means for executing a motion search for a search range of (integer) and generating search results as first to Nth motion vectors,
Means for comparing evaluation function values of the first to Nth motion vectors, selecting a motion vector having the best evaluation function value, and outputting as an evaluation motion vector;
From the external storage device, the search range corresponding to the evaluation motion vector and the search range corresponding to the motion vector for which the vector search unit performs motion search are at least twice the vertical and horizontal directions of the search range. A moving picture encoding program that causes the image data to be read as a predetermined area and to be stored in the internal buffer.   The moving image encoding program according to appendix 7, further causing the computer to function as vector setting means for generating a set motion vector in advance as one of the first to Nth motion vectors. An external storage device for holding a plurality of still images constituting a moving image;
A reference image reading unit that reads image data of a predetermined area of a reference image corresponding to an encoding target image from the external storage device, and stores the image data in an internal buffer;
Using the image data of the predetermined region of the reference image stored in the internal buffer, the first to Nth (N is 2 or more) having different ranges with respect to the encoding target region of the encoding target image. An integer) search range, and a vector search unit that generates search results as first to Nth motion vectors, respectively,
The vector search unit compares the evaluation function values of the first to Nth motion vectors, selects a motion vector having the best evaluation function value, and outputs it as an evaluation motion vector,
The reference image reading unit performs a vertical search of a search range corresponding to the evaluation motion vector and a search range corresponding to a motion vector on which the vector search unit performs a motion search from the external storage device. A moving picture coding system that reads out the image data by using an area obtained by multiplying the horizontal and horizontal at least twice as the predetermined area and stores the read image data in the internal buffer.   The moving image encoding system according to claim 9, further comprising a vector setting unit that generates a set motion vector in advance as one of the first to Nth motion vectors.

Images